If you have servers in the beta system, please login and check that things look okay (and maybe add more servers?). If you don’t have servers in the beta system, please login and try adding some.
If you observe an error, the “trace id” from the HTML or HTTP headers is very helpful.
I’m planning to search the traces and logs next weekend to evaluate how close it is to be ready for the production site. The refactor has been in the works for a long time (with some detours for the infrastructure migration in the spring, various bug fixes, infrastructure tweaks and fixes for the growing monitoring data, etc etc) and I’m very excited to have it in production and be a foundation for the next many years.
There is a redirect loop on selecting “Sign in or create account”:
The page isn’t redirecting properly
Firefox can’t connect to the server at login.ntppool.org
The domain names ntppool.org and beta.grundclock.com shall not appear in the code at all (find them all with global search). The actual domain name shall be a deployment parameter.
Registering a new server as such works, but there’s some oddities/issues:
When registering via name, the name for new servers does not appear on the management page afterwards, or on an individual server’s detail page (it’s showing for legacy servers in both places, though).
Similarly, the zones that are assigned to a server are not showing on the management page (but they do show on an individual server’s detail page).
The score is shown with very high precision on the management page, e.g., Current score: 17.4297568686979 (while it’s shown rounded to a single digit on an individual server’s detail page).
It seems both IPv6 and IPv4 addresses get sorted at the end of the list of servers on the management page, after pre-existing IPv6 servers.
Following the link for the verification yields a “500 - Server Error”, in one example with request id 01M36TRMVBQ97DRF825Z66DDZ8, trace id 82d30042774221a0c7085a55095d73d7.
It used to be possible to schedule a server for deletion, then go through the process of adding it again while not yet fully removed. That was useful, e.g., when a name was to be added to the server retroactively, or the name was to be changed, or when the server’s previous geolocation was wrong, and re-adding the server could fix that when the underlying MaxMind geolocation data had been updated in the meantime. Now, neither of that is possible anymore, refusing the update of the server with an error message, and suggesting to contact support. This forces operators to either wait until the server has actually been deleted, or it forces them to go through support, when they easily could have addressed those cases all by themselves instead. Also, the error message part “Please contact support to restore it.” is confusing, because an operator can actually still cancel the deletion request themselves without the need to contact support.
Sorry for the “me too” post again.
I have the same problem trying to setup as beta monitor as @NTPman . I tried both versions of the ntppool-agent, stable (4.1.6) and test (4.2.0) and both times got the same response:
Sep 23 18:11:10 mail systemd[1]: Started ntppool-agent@test.service - NTP Pool Monitor (test).
Sep 23 18:11:10 mail ntppool-agent[190156]: level=INFO msg="starting ntppool-agent" env=test version="v4.2.0/829b66f (2026-09-21T05:32:30Z, go1.26.8)"
Sep 23 18:11:10 mail ntppool-agent[190156]: level=WARN msg="no API key, please run ntppool-agent setup" env=test cmd="ntppool-agent setup --env test --state-dir '/var/lib/ntppool-agent'" wait_time=33s
Sep 23 18:11:10 mail ntppool-agent[190156]: Setup API key with:
Sep 23 18:11:10 mail ntppool-agent[190156]: ntppool-agent setup --env test --state-dir '/var/lib/ntppool-agent'
Sep 23 18:11:10 mail ntppool-agent[190156]: level=INFO msg="starting AppConfig manager early for file watching" env=test name=""
Sep 23 18:15:02 mail ntppool-agent[190156]: level=WARN msg="no API key, please run ntppool-agent setup" env=test cmd="ntppool-agent setup --env test --state-dir '/var/lib/ntppool-agent'" wait_time=2m19s
Sep 23 18:16:10 mail ntppool-agent[190156]: level=WARN msg="failed to refresh JWT token" env=test appconfig-manager.err="no API key available for JWT token request"
Sep 23 18:17:20 mail ntppool-agent[190156]: level=WARN msg="no API key, please run ntppool-agent setup" env=test cmd="ntppool-agent setup --env test --state-dir '/var/lib/ntppool-agent'" wait_time=3m54s
And like @MagicNTP I can not validate a newly added server on the beta site
500 - Server Error
Ouch! That didn't work, our server hit a bad gear.
(request id 01M37GZ62EVFB11M3XPTV1CW0N, trace id 55ebf438979805540f5bf6715c1f7d4b)
Looking good, glad to see monitor data updated and the Monitors link back on production site.
On the Beta site – several of my servers are showing scores of 19.999999999 – I appreciate the precision, but its probably a bit more than necessary! This only appears on the Manage Servers page (/manage/), not the user page (/a/username) or the individual server page.
Single packet loss in a monitoring batch decreases the score only a tiny bit. Lower display precision would not show this, there would be no way to differentiate this kind of monitor from a really perfect monitoring result.
Do you still get this? I managed to break the production site login (!) for a couple days when fixing the beta site. They should both be working as of now though.
@mopani I spent some time trying to figure out if this was a new behavior, but I think actually it’s an old bug. On most of the UI the scores are rounded to one decimal, but they weren’t there. I fixed that and when there’s a bit more monitoring data in the beta system I will run some analytics to make sure nothing changed to make “slightly less than 20” happen more than before.
Much better now! I stepped over the phase that was failing before, setting up a test monitor. However, the old bug of not offering the nearest airport (~10km) to my monitoring location is there, or feature of free entering the airport code is still missing.
Registration registering the name properly now, verification links can be obtained, ordering on management page now as it was, scores with reasonable and useful rounding again, thanks!
Zones still missing. And I personally find the new feature of blocking re-registration of servers scheduled for deletiom unnecessary, and actually unfortunate as it needlessly (unless there is a strict technical need) restricts useful self-help options for server operators.
Small observation/nit: The deletion time for a server is now given including the hour:
This server is scheduled for deletion on 2026-09-30T00:00:00Z.
Could be considered another rounding issue. If intentional for clarity of when exactly the deletion is going to happen, then maybe the “on” in that sentence should be replaced by “at”.
However, the old bug of not offering the nearest airport (~10km) to my monitoring location is there, or feature of free entering the airport code is still missing.
Is your nearest airport in another country from the GeoIP data? The system filters those out as options. While just about everyone contributing are thoughtful and trying to have the system work as well as possible, system has a lot of little guardrails to steer people the right way if they aren’t as thoughtful or understanding of how the system works.
If I am guessing right about the country, the guardrail is there in part to have the monitoring data and diagnostics not be confusing for everyone else and in part because when I was testing it was a weird experience when I simulated an IP in Copenhagen and then getting country / airport codes in Sweden (or Malaysian ones with a Singapore IP, that sort of thing).
The system also only does IATA codes, so if the airport only has a LFLI code it won’t show up.
If you’d rather email and privately message the details to help me understand, then please do that!
Really what I wanted was names based on the UN/LOCODE codes, but I couldn’t find a reasonable API for looking those up by a location.
Zones still missing. And I personally find the new feature of blocking re-registration of servers scheduled for deletiom unnecessary, and actually unfortunate as it needlessly (unless there is a strict technical need) restricts useful self-help options for server operators.
Should be fixed now; also the date formatting (a few more tweaks coming up that didn’t make it to the build I made earlier in the afternoon). Thank you!
